I tried to run 'nodetool compact' (or with keyspace and cfname), seemed not woking. The command hung there and nothing happened. But I noticed that there was about more 30000+ pending tasks when using 'nodetool compactionsats'
On Tue, May 6, 2014 at 1:45 AM, Robert Coli <rc...@eventbrite.com> wrote: > On Mon, May 5, 2014 at 12:24 AM, Daniel Chia <danc...@coursera.org> wrote: > >> If you want to trigger a conversion to LCS of the on-disk sstables, doing >> "nodetool compact <cf>" should achieve what you want. >> > > For more detail and a caveat : > > https://issues.apache.org/jira/browse/CASSANDRA-6092 > " > The workaround to call "nodetool compact" only works when there is at > least some activity, causing there to be more than 1 sstable. I've just > tested with a cluster that was not receiving any reads nor writes. In that > case calling "nodetool compact" does not do anything ... > " > > =Rob > >
